Selectivity and Efficacy of Herbicides Dimethachlor and Pethoxamid in Rocket Crop

Plant Protection Science, 56, 2020 (4): 305–316 Original Paper https://doi.org/10.17221/93/2020-PPS

Selectivity and efficacy of herbicides dimethachlor and pethoxamid in rocket crop

Ivana Doležalová1, Irena Petrželová1, Martin Duchoslav2*

1Department of Genetic Resources of Vegetables, Medicinal and Special Plants of the Crop Research Institute in Olomouc, Olomouc, Czech Republic 2Department of Botany, Faculty of Science, Palacký University, Olomouc, Czech Republic *Corresponding author: [email protected]

Citation: Doležalová I., Petrželová I., Duchoslav M. (2020): Selectivity and efficacy of herbicides dimethachlor and pethoxamid in rocket crop. Plant. Protect. Sci., 56: 305–316.

Abstract: Field experiments were conducted to evaluate the efficacy, selectivity and health harmlessness of four -ap plication rates of two pre-emergent herbicides (pethoxamid and dimethachlor) in the rocket [Eruca vesicaria (L.) Ca- vanilles)]. Pethoxamid was found to be less efficient on the total weed density (efficacy 86.0–93.3%) in comparison with the effect of dimethachlor (94.9–95.8%). Dimethachlor expressed an excellent efficacy Echinochloaon crus-galli (L.) P. Beauvois, Portulaca oleracea Linnaeus, Amaranthus retroflexus Linnaeus, Lamium purpureum Linnaeus, and Ve- ronica persica Poiret from the lowest tested application rate (800 g/ha). Pethoxamid showed an excellent efficacy on E. crus-galli, Lamium purpureum, Lamium amplexicaule Linnaeus, V. persica, and P. oleracea. In higher application rates, pethoxamid controlled Chenopodium polyspermum Linnaeus and Chenopodium album Linnaeus. In contrast to mostly negative effects of dimethachlor, pethoxamid showed either no effects or positive ones on the rocket yield. Residues of both herbicides in the harvested product were always below a 'default limit', which is the baseline maximum residue level for food. The selectivity of pethoxamid at an application rate of 960 g/ha was good, herbicide residues in the rocket were not detected and the yield of the rocket increased.

Keywords: Eruca vesicaria (L.) Cav.; maximum residue levels (MRLs); phytotoxicity; weed control

Rocket salad, arugula, or roquette [Eruca vesicaria Weeds are a major threat to most crops. In addi- (L.) Cavanilles)], belonging to the Brassica family tion to the competition effect, weeds present in the (Brassicaceae), is one of the oldest crops tradition- field at harvest time can contaminate the rocket, es- ally grown in Southern Europe and Western Asia. pecially when it is harvested mechanically (Holm- Rocket is traditional in Mediterranean cuisine as strom 2008). Any admixture in the harvested prod- a leafy green, but in recent decades it has been be- uct may thus reduce product quality. Concerning coming more popular in Central Europe and in the the biological control of weeds, there are no effective United States. In the kitchen, the rocket is primar- methods available for the rocket crop. Hand weeding ily used fresh in salads, pasta dishes, or cooked and and mechanical cultivation are common practices eaten like spinach. Rocket is an annual, cool-season but time-consuming and costly. To select an appro- crop that flowers under long days and high tempera- priate herbicide, it is necessary to know which weed tures. It is quick growing, being especially produc- species occur in the field. With respect to ensuring tive in spring, early summer, and autumn (Padulosi both effectiveness on the weeds and selectivity to 1995; Doležalová et al. 2013). crop, pre-emergent herbicides are more promising,

Supported by the Project No. RO0418 (MA) and by the institutional support from Palacký University.

305 Original Paper Plant Protection Science, 56, 2020 (4): 305–316

https://doi.org/10.17221/93/2020-PPS because they control germinating weeds. Bensulide selected pesticides throughout the European Union is the pre-emergent herbicide most commonly used by the Regulation (EC) No. 396/2005. According to in a rocket in Arizona and New Jersey (USA); it is this regulation, MRLs are the highest levels of resi- effective against grass weeds and also controls some dues expected to be in the food when a pesticide is small-seeded dicot weeds. Sethoxydim and other used according to the authorized agricultural prac- ACCase inhibitors are post-emergent herbicides tices. The regulation also established a general lim- that control only grasses (Dimson 2001). Another it that applies in cases when no specific MRL has option is to use non-selective herbicides with active been defined for a particular crop (a 'default limit' of substances such as carfentrazone-ethyl, paraquat 0.01 mg/kg) as it is in the case of the rocket. dichloride, pelargonic acid, pyraflufen-ethyl, and The present study aims (i) to evaluate the efficacy glyphosate to control weeds in the field before the of four application rates of herbicides dimethachlor rocket sowing (Dimson 2001; Holmstrom 2008). and pethoxamid, (ii) to compare the impact (i.e. the The problematics of the chemical control of weeds possible toxicity and influence on plant vitality and in the rocket crop have not been studied yet in the yield) of these herbicides on the rocket, and (iii) to temperate climatic conditions of Central Europe. evaluate the presence/absence of herbicide residues Since there is no pesticide registered for the rocket in the harvested product. in the Czech Republic (Anonymous 2014), two herbicides which are registered in the Czech Republic for MATERIAL AND METHODS the related species (i.e. rapeseed and various crucifer- ous vegetables), were considered to test their efficacy, Plant material and herbicide selection. The seed selectivity, and health harmlessness in the rocket. of rocket salad originating from the seed company Both selected herbicides belong to chloroacetamides, Semo Ltd. (Smržice, Czech Republic) was used in which are categorized into the group 15 according the experiments. On the basis of a preliminary ex- to Herbicide Resistance Action Committee (HRAC) periment in 2014 (Doležalová et al. 2014), two pre- classification (HRAC 2010). They areabsorbed by emergent herbicides (Teridox 500 EC, Syngenta new shoots of seedlings and roots and a mode of their Czech Inc., active ingredient dimethachlor; Somero action is through the inhibition of lipid synthesis 600 SC, Arysta LifeScience Czech Inc., active ingre- and cell division (VLCFAs inhibition), which inhib- dient pethoxamid), which showed minimal injury its weed germination (Böger et al. 2000; HRAC 2010; of rocket and should control a broad spectrum of Yang et al. 2010; PPDB 2020). The effect of pethoxam- weeds, were selected for the experiments. id is systemic (PPDB 2020). Pethoxamid can be safely Design of experiments. The experiments were applied pre- and early postemergently in a wide range conducted in the experimental fields on the periph- of arable and horticultural crops and controls a broad ery of the city of Olomouc (the Czech Republic, spectrum of grass and broad-leaved weeds effectively N 49°34'21.37", E 17°17'1.21", 222 m a.s.l.) in two sea- (Hunt et al. 2015; PPDB 2020). Dimethachlor is used sons. The fields of the Crop Research Institute (the year pre-emergently to control most annual grasses and 2015) and of the Central Institute for Supervising and broad-leaved weeds (PPDB 2020). Testing in Agriculture (the year 2016) were used for Despite the benefits of pesticides for crop yields the experiments. The previous crops were peas with and their relevance for the economy, intensive and soybeans and potatoes, respectively. The first field was widespread pesticide use raises serious environ- an example of a heavily weeded one, while the second mental and health concerns (Damalas & Elefthero- one was an adequately weeded field. Both fields were horinos 2011; Nicolopoulou-Stamati et al. 2016). without previous treatment with any non-selective or Therefore, for each newly registered application of other herbicide. The experimental fields lie on alluvial a pesticide for protection of a particular crop, it is sediments; the soils are fluvisols, with a sandy-clay or necessary to be very thorough in setting an applica- clay soil texture (Tomášek 2015). The soil pH was 6.7. tion rate that can be recommended to users as harm- During the preparation of the experimental plots, the less to health, especially in leafy vegetables which soil was fertilized with LOVOFERT NPK 15-15-15 (Lo- have a short growing season. In order to limit the vochemie Ltd., Czech Republic) in a dose of 50 kg/ha risks of presence of pesticide residues in foodstuffs, (i.e., 7.5 kg/ha of each of N, P2O5, and K2O). Before the maximum residue levels (MRLs) in or on food the sowing (14th April 2015 and 20 th April 2016) plots and feed of plant and animal origin were defined for were cultivated using conventional tillage. The soil

306 Plant Protection Science, 56, 2020 (4): 305–316 Original Paper https://doi.org/10.17221/93/2020-PPS was sufficiently moist before the sowing and the -ap pethoxamid 0.8 RD, RD, and 1.2 RD. The analyses plication of herbicides. The seed volume was about were carried out at the National Reference Labo- 110 seeds per m2, the sowing depth 1.5–2.0 cm, and ratory of the Central Institute for Supervising and the inter-line spacing 25 cm. The seeds were sown on Testing in Agriculture (Brno, Czech Republic) using plots with an area of 13.5 m2 (width 3 m, length 4.5 m, the standard LC-MS method (Anonymous 2020) on 12 rows) using a three-row Stanhay Robin small seed an instrument UPLC-MS/MS Xevo (Waters, USA). drill with belt metering units (Stanhay Robin, UK). Climatic conditions in experimental fields.The The herbicides were tested according to the Euro- course of the weather during the growing seasons pean and Mediterranean Plant Protection Organiza- was measured at a meteorological station located tion (EPPO) standard PP 1/49 (3) for the evaluation near the experimental field. The experimental peri- of their efficacy in Brassica oil crops (OEPP/EPPO ods from the sowing to the assessment and harvest 2007). The hectare application rate recommended lasted 51 days in 2015 and 50 in 2016. The average (RD) for rapeseed is 1 000 g/ha of dimethachlor and temperatures at the locality during those periods 1200 g/ha of pethoxamid. Four application rates were 15.8 °C in 2015 and 14.5 °C in 2016, and the to- (0.8 RD; RD; 1.2 RD; 2.0 RD) of each of the two se- tal precipitation was 63.4 mm in 2015 and 52.5 mm lected herbicides and pure water as control were in 2016. Detailed meteorological data during the tested with the water volume 300 L/ha. The experi- course of the experiments is given in the ESM. ments were realized in four replications in randomly Data analyses. Partial Principal Component patterned plots. Analysis (pPCA; Legendre and Legendre 2012) was The herbicides were applied with a HEGE 32 trial carried out to assess the main gradients in the com- sprayer with a HEGE 76 tool carrier (Hege Maschin- position of the weed species in the plots treated with en GmbH, Germany) one day after the sowing of the different herbicide treatments. The species data rep- rocket. The boom arm of the sprayer was equipped resented the logarithmically transformed number of with six flat fan nozzles (Albuz API 110 02; SOLCERA individuals [log(x + 1)]. To control for the possible Advanced Materials, France) with the height of the effect of different seasons on the species composi- nozzles being 50 cm and the swath width of the spray tion, the year was used as a covariate (block) in the 3 m. The application was performed at a travel speed analysis. Each treatment group (herbicide × applica- of 2.6 km/h, a flow rate of 0.066 L/s, and a nozzle tion rate) was visualized in an ordination diagram. pressure of 0.2 MPa. After the application, the herbi- Visualization of the total number of weed individu- cides were not incorporated into the soil. Because of als and the number of weed taxa per plot in the ordi- the quite low frequency and/or quantity of the rain- nation space was based on contour plots displaying fall during the experimental periods (see Electronic isolines of a response surface fitted with generalized Supplementary Material – ESM), the experimental linear models (GLM). Model complexity was evalu- fields were irrigated as needed using above-ground ated using the Akaike Information Criterion statistic sprinklers. The numbers of individuals of weed spe- (AIC; Šmilauer & Lepš 2014). cies occurring in the experimental and control plots Constrained ordination (partial redundancy anal- were evaluated based on a 1m2 area in the center of ysis, pRDA; Šmilauer & Lepš 2014) was used to test each plot at two points in time (five and seven weeks the direct effect of herbicide treatment (irrespective after application). The yield of rocket leaves at harvest of herbicide type and application rate) on the spe- maturity (7 weeks after application; 3 June 2015 and cies composition within the plots. The significance 8 June 2016) harvested from the 1m2 areas was as- of the effects of herbicide treatment vs. water-treat- sessed. In each 1m2 plot, the rocket plants were first ed control was tested by a Monte Carlo permutation counted, then harvested, and their fresh biomass was test with 999 permutations (MC test) and the year divided into consumable and inconsumable parts, was used as a covariate (block) in the analysis. Sec- which were immediately weighed. ond constrained ordination (pRDA) was performed The content of residues of the herbicides (i.e. di- with identical settings to the former analysis but on methachlor and pethoxamid) was analysed at the a reduced dataset containing only herbicide-treated time of the rocket maturity in the leaves harvested plots. The most parsimonious model was selected from the treatments with the lowest recorded phyto- according to the backward elimination of insignifi- toxicity of the herbicides that were tested for rocket cant effects using an MC test. To find out the weed plants, i.e. for dimethachlor 0.8 RD and RD, and for species that were either suppressed or enhanced

307 Original Paper Plant Protection Science, 56, 2020 (4): 305–316

https://doi.org/10.17221/93/2020-PPS by the application of the respective herbicide or untreated control plots along the first ordination axis specific application rate, a t-value biplot was used. (capturing 38.9% of the total variation; Figure 1A). Because of the categorical nature of the predictors The scores of plots along the second ordination axis we utilized, the effect of each application rate was (8.3%) partly reflect the differences in the composi- compared to RD. Specifically, we used Van Dobben tion and abundance of the weed species in the con- circles, which allow easier interpretation of a t-value trol plots between the two experimental years. The biplot (Lepš & Šmilauer 2014). The Canoco software control plots featured multispecies weed composi- (Version 5.10) (ter Braak 5.10; Šmilauer 2012) was tion (7–17 weed taxa per 1 m2 plot; Figure 1B and used for the above-mentioned analyses. C) and dense weed populations (Figures 1D and 3) The effects of the herbicide and application rate on in comparison with the herbicide-treated plots. In the total number of weed plants per m2, number of the control plots, several annual weeds such as Che- plants of most abundant weed species per m2, and the nopodium polyspermum Linnaeus, Chenopodium yield of rocket (number of rocket plants, the weight album Linnaeus, Amaranthus retroflexusLinnaeus, of fresh consumable and inconsumable parts per m2) Echinochloa crus-galli (L.) P. Beauv., Lamium pur- were analysed separately for each year using linear pureum Linnaeus, Lamium amplexicaule Linnaeus, models (LM) using REML estimations in NCSS (ver- Veronica persica Poiret, and Portulaca oleracea Lin- sion 9). The herbicide (pethoxamid, dimethachlor) naeus were common and usually formed dense pop- and application rate (with five levels: Control, 0.8 RD, ulations. On the other hand, the herbicide-treated RD, 1.2 RD, and 2.0 RD) and their interaction were plots were less diverse concerning species (less than considered as fixed-effect predictors. The square root five taxa per 1 m2 plot) and no weed species tended of the weed density seven weeks after the applica- to prefer herbicide-treated plots. The differences tion of the herbicide was considered as a covariate in in the weed species composition between the her- the analysis of the yield of the rocket. Efficacy (in %; bicide-treated and untreated plots were significant range 0–100%) of tested herbicide treatments on (pRDA; MC test, pseudoF = 97.6, P = 0.002, 38.3% of selected weeds was calculated from estimated mean the adjusted explained variance). weed densities, comparing results for respective ap- Detailed inspection of the ordination pPCA dia- plication rate to control, and interpreted according to gram showed more extreme left positions of (i) the Jursík et al. (2015). Only weed species with sufficient plots treated with dimethachlor than those treated densities (> 2 plants per 1 m2) in the control plots with pethoxamid, and (ii) higher than lower ap- were analysed for herbicide efficacy. For multiple plication rate plots, indicating a higher efficacy of comparisons, a Bonferroni test was used. dimethachlor and higher herbicide rates in general For the statistical analysis of the herbicide resi- against weeds (Figure 1A). Therefore, we further dues, nondetects data analysis (Hensel 2005) was focused on a reduced dataset containing just the calculated in NCSS. Specifically, the contents of herbicide-treated plots. Simplification of the model residues among three treatments were compared resulted in a final pRDA model with just the main for each herbicide. A log-rank test was used for test- effects of the herbicide (pseudoF = 18.0, P = 0.002, ing the null hypotheses, that the distribution func- 12.2% of the adjusted explained variance) and appli- tions of two or more populations produced using cation rate (pseudoF = 6.4, P = 0.002, 11.4%) being the Kaplan-Meier product-limit estimator are equal. significant. Dimethachlor was found to be more ef- When the overall test was significant, multiple pair- fective in the control of many weeds than pethoxa- wise log-rank tests were calculated for each pair of mid (e.g. E. crus-galli, Persicaria sp., and A. retrof- treatments to find out which application rates dif- lexus; Figure 2A). Comparing different application fered from each other. rates with the RD also showed that an application rate below the RD controlled some weeds (E. crus- RESULTS -galli and A. retroflexus) less efficiently than the RD (Figure 2B). 1.2 RD did not differ in its effect from Effect of different application rates of two pre- the RD (Figure 2C). On the other hand, 2.0 RD con- emergent herbicides on weed species composi- trolled several weed taxa (e.g. Lamium purpureum, tion and density in rocket stands. Partial PCA of L. amplexicaule, Persicaria sp., Thlaspi arvense L., the composition of the weed taxa (for detailed data A. retroflexus, E. crus-galli, etc.) more efficiently see ESM) differentiated the herbicide-treated and than the RD (Figure 2D).

308 Plant Protection Science, 56, 2020 (4): 305–316 Original Paper https://doi.org/10.17221/93/2020-PPS 3 1.0 0.8 PortOler 7 (A) (B) (C) 9 11 13 5 15 17 ChnPol, ChnAlb GalnParv PoaAnn AmarRetr –1.0 –1.0 2.0 Control MercAnnu PerscSp EchnCrus DIM 0.8RD (D)

DIM RD ThlArv 1.0 DIM 1.2RD StelMedi 500 1 000 DIM 2.0RD GaleTetr AnagArvn PET 0.8RD 1 500 PET RD SoncArvn Lami PET 1.2RD TripMart

–3 PET 2.0RD –0.4 VernPers –2 3 0.0 1.0 2 000 –1.0 –1.0 2.0 Figure 1. Partial principal component analysis (pPCA; first and second ordination axes) of weed composition in herbicide-treated and untreated control (water-treated) plots with experiments in two consecutive years treated as covariates (blocks) (A) ordination diagram of plots with visually distinguished treatments (circles : DIM – dimethachlor; squares = PET – pethoxamid; crosses : untreated control; colouring means different doses : RD – application rate recommended for rape, 0.8, 1.2 and 2.0 are multiples of RD). (B) Ordination diagram of weed species; the taxa in blue are significantly negatively affected by any herbicide treatment, on the evidence of t-value biplot analysis. (C) Contour plots (GLM-based, predictors CaseR.1+CaseR.2, distribution quasi-Poisson, link function log, fitted model deviance 188.1, with 155 residual dfs, null model deviance 311.3, with 159 residual dfs, dispersion parameter 1.2, model AIC 776.3, model test F = 25.6, P < 0.001) of species richness within plots in the ordination space. (D) Contour plots (GLM-based, predictors CaseR.1+CaseR.2, distribution Gamma, link function log, fitted model deviance 204.45, with 155 residual dfs, null model deviance 482.45, with 159 residual dfs, model AIC 1631.09, model test F = 65.1, P < 0.001) of weed density within plots in the ordination space. AmarRetr – A. retroflexus; AnagArvn – A. arvensis; ChnPol, ChnAlb – mainly Chenopodium polyspermum L., sporadi- cally Chenopodium album L.; CirsArvn – Cirsium arvense (L.) Scop.; EchnCrus – E. crus-galli; EuphHeli – Euphorbia helioscopia L.; GaleTetr – Galeopsis tetrahit L.; GalnParv – Galinsonga parviflora Cav.; Lami – Lamium purpureum, L. amplexicaule; MercAnnu – Mecurialis annua L.; PerscSp – Persicaria sp. div., mainly P. maculosa; PoaAnn – Poa annua L.; PortOler – P. oleracea; SoncArvn – Sonchus arvensis L.; StelMedi – Stellaria media (L.) Vill.; ThlArv – T. arvense; TripMart – Tripleurospermum maritimum (L.) W.D.J Koch.; VernPers – V. persica The total densities of weeds seven weeks after (b) despite a trend of decreasing weed densities with the application of the herbicide (Figure 3) were sig- an increasing herbicide application rate, only 0.8 RD nificantly affected by the herbicide (2015:F = 13.3, and RD resulted in statistically less effective sup- df = 1, 24.6, P < 0.001; 2016: F = 10.7, df = 1, 20.5, pression of weed densities compared to the plots P = 0.004) and application rate (2015: F = 127.4, treated with 2.0 RD in 2015 (Figure 3A); (c) except df = 4, 12.4, P < 0.001; 2016: F = 36.6, df = 4, 12.1, for 1.2 RD, pethoxamid and dimethachlor did not P < 0.001), while the effect of herbicide × applica- differ statistically in their effects on the weed density tion rate interaction was significant in 2016 (F = 3.6, in 2016 (Figure 3B). df = 4, 12.1, P = 0.038) but not in 2015 (F = 0.4, df = 4, The efficacy of the tested herbicides on weeds was 12.4, P = 0.825). Efficacy of dimethachlor on weeds studied in detail for the eight most abundant taxa (irrespective of concentration) was good to very good (Table 1). Dimethachlor fully controlled Echinoch- (2015: 94.9%; 2016: 95.8%), while the application of loa crus-galii, Veronica persica, and Amaranthus pethoxamid was slightly less effective (2015: 93.3%; retroflexus (in 2016) from the lowest application 2016: 86.0%). Detailed comparisons of the effects rate. Control of Amaranthus retroflexus (in 2015) of the application of different herbicide application and Portulaca oleracea (90–99% and 91–97%, re- rates on weed densities showed that (a) the weed spectively) was good to very good. Only the highest densities in the untreated plots were significantly tested application rate controlled Thlaspi arvense higher (P ≤ 0.05) than the densities in every plot acceptably. Chenopodium polyspermum was con- treated with herbicide in both years (Figure 3A, B); trolled effectively only by the two highest application

309 Original Paper Plant Protection Science, 56, 2020 (4): 305–316

https://doi.org/10.17221/93/2020-PPS 1 (A) (B) (C) EuphHeli (D) VernPers AnagArvn CirsArvn AmarRetr PerscSp ThlArv PerscSp EchnCrus Lami Lami PerscSp EchnCrus Lami EchnCrus EuphHeli ChnPol, ChnAlb ChnPol, ChnAlb ThlArv AmarRetr AmarRetr PortOler –1 –1 1–1 1 –1 1–1 1 Figure 2. The t-value biplots and Van Dobben circles that summarize the t-statistics of the regression coefficients representing the relation of each weed species in a constrained ordination (pRDA) to the predictors (herbicide, application rate) (A) effect of pethoxamid (PET) compared with dimethachlor (DIM) as predictors, irrespective of application rate; (B–D) Effect of each application rate (0.8 RD, 1.2 RD, 2.0 RD as predictors) compared to a RD (triangle in the centre of the diagrams), irrespective of the herbicide used. Van Dobben circles were plotted in pairs reflecting positive (coloured red) or negative effects (coloured blue) of selected predictor variables on weed species. When a species' arrowhead ends within a positive response circle (plotted in red), the effect of the predictor variable upon that response variable is judged significant, because the t-value statistic from a (multiple) regression is predicted to have a value greater than two. When the arrowhead falls within the negative (blue) response circle, a significant negative response (based on a t-value less than –2) is predicted (ter Braak & Šmilauer 2012). Only the weed species significantly affected by the respective predictor are reported in figures. For species abbreviations see Figure 1 rates under low weed infestation (2016). Dimetha- nopodium polyspermum. Control of Thlapsi arvense chlor controlled C. polyspermum insufficiently -un and Persicaria sp. was insufficient. Efficacy of both der intensive weed infestation (2015). Control of tested herbicides was very good (96–100%) on La- Persicaria sp. was insufficient. mium purpureum and L. amplexicaule, even from The highest tested application rate of pethoxamid the lowest tested application rate. Both herbicides fully controlled Portulaca oleracea, Veronica per- fully controlled Lamium with 2.0 RD. sica, Amaranthus retroflexus, and Echinochloa crus- Effect of different application rates of two pre- galii. The efficacy of other application rates on the emergent herbicides on yield parameters of rock- mentioned species was good to very good. Only the et stands. The analysis of the production traits of highest application rate acceptably controlled Che- the rocket showed marked differences between the (A) (B) 1 000 1 000

Figure 3. Weed densities (mean + 2 plot)

2 SE; plants per 1 m ) in the un- DIM treated control plots and plots 100 100 treated with different application PET rates (0.8, 1.0, 1.2 and 2.0 times the recommended application rate, RD) of two herbicides (dimethachlor (DIM) and pethoxamid (PET)) in (A) 2015 and 10 10 (B) 2016 Note the log-scale of weed density. Significant differences Weed density (weed plants/1 m Weed between groups (Bonferroni test) are marked (* 0.01 < P ≤ 0.05, ** 1 1 0.001 < P ≤ 0.01, *** P ≤ 0.001). Control 0.8 RD RD 1.2 RD 2.0 RD Control 0.8 RD RD 1.2 RD 2.0 RD For details, including the results Application rate Application rate of LM, see the text

310 Plant Protection Science, 56, 2020 (4): 305–316 Original Paper https://doi.org/10.17221/93/2020-PPS

a a a a a a a a 91 97 97 96 99 100 100 100 2016 -value in -value in 52 (54) P oleracea Portulaca 4.4 (0.051) 1.1 (0.393) 42.4 (< 0.001) a a a a a a a a 97 98 98 100 100 100 100 100 2015 56 (41) persica Veronica Veronica 0.5 (0.511) 0.7 (0.592) 68.2 (< 0.001) ) and respective F a a a a a a a a

90 97 100 100 100 100 100 100 2016 10 (5) 3.2 (0.085) 0.8 (0.510) 62.2 (< 0.001) ≤ 0.05 (Bonferroni test); RD – application ≤ 0.05 (Bonferroni test); RD – application P retroflexus Amaranthus Amaranthus a a a a a a a a 90 97 97 89 89 85 99 97 2015 66 (41) 1.6 (0.229) 7.7 (0.002) 1.2 (0.370)

a a a a a b ab ab 43 50 51 50 48 90 79 70 2015 Thlaspi arvense 51 (38) 1.6 (0.221) 8.3 (0.022) 3.5 (0.043) , a a a a a a a a spp. 63 74 56 96 81 81 89

100 7 (4) 2016 2.6 (0.119) 0.9 (0.513) 15.8 (0.001) mainly mainly a a b b b b a a 0 0 36 69 17 69 74 92 C. polyspermum 2015 Chenopodium Chenopodium 13 (16) 7.0 (0.014) 3.0 (0.063) 0.1 (0.980) a a a a a b ab ab

73 91 91 100 100 100 100 100 2016 20 (12) 2.3 (0.083) 28.1 (< 0.001) 43.7 (< 0.001) crus-galli Echinochloa b b b a a a a ab 99 94 97 98 99 100 100 100 2015 568 (141) 85.4 (< 0.001) 12.7 (< 0.001) 259.5 (< 0.001)

0 45 55 51 69 27 52 67 32 (33) Persicaria 0.4 (0.520) 2.0 (0.160) 0.7 (0.640) P. maculosa P. spp., mainly L. , b b a a a a a a 97 98 99 96 97 98 100 100 2015 169 (90) Lamium 1.6 (0.221) 0.5 (0.737) amplexicaule 64.6 (< 0.001) purpureum

0.8 1.0 1.2 2.0 0.8 1.0 1.2 2.0 rate App. App.

) 2 Herbicide Dimethachlor Dimethachlor Dimethachlor Dimethachlor Pethoxamid Pethoxamid Pethoxamid Pethoxamid Weed density density Weed plots in control (ind./m Herbicide Application rate × Herbicide rate application Table 1. Efficacy (in %) of the tested application rates (0.8, 1.0, 1.2, and 2.0 of RD) of two herbicides (dimethachlor, pethoxamid) on selectedon weedpethoxamid) species shortly (dimethachlor, herbicides of 2.0 two of and RD) 1.2, 1.0, (0.8, rates Efficacy 1. application tested the of %) (in Table rocket harvestbefore the (seven weeks after sowing) effectson weed interaction The and their tested density (mean LM; rate were of herbicide,by the and SD) test value application the ( reportedparentheses are each test; for efficacy not significantly at values are same letter different with withinthe a column rate recommended for a rapeseed is 1000 g/ha of dimethachlor and 1200 g/ha of pethoxamid; four application rates are multiplies of RDof recommended multiplies rapeseedRD;a (0.8 are for RD; rate rates RD;1.2 RD)2.0 application four pethoxamid; of g/ha 1200 and is dimethachlor of g/ha 1000

311 Original Paper Plant Protection Science, 56, 2020 (4): 305–316

https://doi.org/10.17221/93/2020-PPS experimental years. In 2015, rocket plants were pro- case of 2.0 RD), while only the effect of the 2.0 RD nouncedly smaller and their leaves were yellowish or of pethoxamid reduced the density of the rocket sig- slightly reddish in the control plots, while the plants nificantly. The weed density had no significant effect were much larger and with bigger leaves in the her- on the rocket yield at all. The application of 0.8 RD, bicide-treated plots. No significant effects of weed RD, and 1.2 RD of dimethachlor, significantly re- density, herbicide, application rate, and their interac- duced the weights of both the consumable and in- tion on the number of rocket plants and the weight consumable parts of the rocket. The application of of consumable parts per m2 were found (Table 2). 2.0 RD of pethoxamid caused a drop in the weight of Only one effect was found significant in the case of inconsumable parts in comparison with other tested inconsumable parts: the application of dimethachlor application rates of pethoxamid. The most extreme significantly reduced the weight of inconsumable drop in the weights of both consumable and incon- parts in comparison with the application of pethoxa- sumable parts of the rocket was caused by the appli- mid. The data also showed a marginally significant cation of 2.0 RD of dimethachlor (Table 2, Figure 4). trend: the application of herbicide in any applica- Herbicide residue assessment. The content tion rate increased the weight of consumable rocket of residues of the herbicides was analysed in the parts in comparison with the untreated control (con- leaves at the time of the rocket maturity. Concern- trast control vs. the average of four application rates: ing dimethachlor, all measurements fell below the F = 4.38, P = 0.072; Figure 4). detection limit (0.003 mg/kg). On the other hand In 2016, the application of dimethachlor in any (overall log-rank test, χ2 = 10.3, df = 2, P = 0.006), used application rate caused a strong reduction in rocket leaves sampled in plots treated with the RD the density of the rocket (even more drastic in the of pethoxamid had significantly higher concentra-

160 600 200 n.s. ) n.s. ) 2 PET > DIM 2 500 120 150 400

80 300 100 200 40

No. of rocket plant 50 Consumable partsConsumable (g/m 100 Inconsumable partsInconsumable (g/m

0 0 0 160 600 200 PET ) ) 2 2 500 DIM a 120 a 150 a ab ab ab 400 ab ab a a b ab b a 80 300 a a 100 b a 200 ab b 40 c c b 50 No. of rocket plant c c c

Consumable partsConsumable (g/m 100 c Inconsumable partsInconsumable (g/m d c d 0 0 0 Control 0.8 RD RD 1.2 RD 2.0 RD Control 0.8 RD RD 1.2 RD 2.0 RD Control 0.8 RD RD 1.2 RD 2.0 RD Application rate Application rate Application rate

Figure 4. Production traits (mean + SE) of rocket (number of plants, weight of consumable and inconsumable parts per m2) in plots treated with two herbicides (dimethachlor (DIM) and pethoxamid (PET)) in four application rates (0.8 RD, RD, 1.2 RD and 2.0 RD) in comparison with the untreated control (Control) Upper row: results of 2015 experiment; lower row: results of 2016 experiment; significant differences between groups (Bonferroni test at P ≤ 0.05) are marked by different letters, separately for each trait and year; n.s – no significant differ- ence; PET > DIM – effect of pethoxamid greater than effect of dimethachlor at P ≤ 0.05

312 Plant Protection Science, 56, 2020 (4): 305–316 Original Paper https://doi.org/10.17221/93/2020-PPS

Table 2. Effect of weed density, herbicide, application rate and their interaction on the yield parameters of rocket in the two experimental years (2015, 2016) analyzed using linear models

2015 2016 Term F dfG dfE P F dfG dfE P Number of rocket plants per 1 m2 $ Weed density$ 0.37 1 12.9 0.552 2.69 1 8.8 0.136 Application rate 0.54 4 13.9 0.709 9.63 4 13.1 < 0.001 Herbicide 0.59 1 26.3 0.449 261.53 1 28.4 < 0.001 Application rate × herbicide 0.44 4 12.0 0.775 14.67 4 12.0 < 0.001 Weight of consumable parts (g/m2)$$ Weed density$ 0.07 1 17.9 0.802 0.00 1 13.3 0.99 Application rate 0.17 4 14.7 0.952 4.62 4 11.7 0.017 Herbicide 0.28 1 23.4 0.604 45.25 1 11.73 < 0.001 Application rate × herbicide 0.14 4 12.1 0.963 4.65 4 11.0 0.019 Weight of inconsumable parts (g/m2)$$ Weed density$ 2.05 1 13.2 0.176 2.26 1 6.8 0.177 Application rate 0.77 4 13.6 0.563 7.04 4 11.8 0.004 Herbicide 6.24 1 22.2 0.020 172.77 1 20.1 < 0.001 Application rate × herbicide 1.14 4 11.2 0.385 10.15 4 11.7 0.001 $ square root transformed; $$ log(x + 1) transformed tions of herbicide (median and its 95% confidence to full control) that is in line with PPDB (2020) and interval: 0.003, 0.002–0.007 mg/kg) than those treat- Veronica persica (full control). This herbicide is per- ed with 0.8 RD (χ2 = 6.0, df = 1, P = 0.015) or un- haps also effective againstGalinsoga parviflora, be- treated with pethoxamid (χ2 = 4.5, df = 1, P = 0.034), cause it was recorded in the untreated plots but not where all the measurements fell below the detection in the adjacent experimental plots. But we cannot limit (comparison Control-0.8 RD: χ2 = 0.0, df = 1, confirm this as we do not know whether it was due P = 1.000). However, the contents of the residues of to the efficacy of the herbicide or the real absence of both herbicides in the rocket leaves treated with dif- the species from the treated plots. Low to acceptable ferent application rates of herbicide were all below control was also recorded against Thlaspi arvense and the 'default limit' of 0.01 mg/kg. Cirsium arvense. Our experiments did not show efficacy against Persicaria sp. div. and Chenopodium DISCUSSION sp. div. Dimethachlor should be efficient also against Alopecurus myosuroides Huds., Apera spica-venti (L.) In both experimental years, clear effects of both P. Beauv., Bromus L., Poa L., Stellaria media (L.) Vill. tested herbicides against weedy species were record- and Tripleurospermum Sch. Bip. (PPDB 2020). ed compared to the untreated controls (Figure 1A). Pethoxamid was less effective against the moni- Higher application rates of herbicides were more ef- tored weed species in comparison with the effect fective than lower ones (Figure 1A, 2A and 3). Di- of dimethachlor. However, in comparison to the methachlor showed better weed control (efficacy on untreated plots, the reduction in the total numbers total weed spectrum 94.9–95.8%). Excellent efficacy of weed plants was also pronounced (86.0–93.3%), (good to full control) of dimethachlor from the low- which could be considered as an acceptable to good est tested application rates was recorded for Echino- weed control on fields with lower weed infestation. chloa crus-galli, Portulaca oleracea, and Amaran- Nevertheless, there were mostly no obvious differ- thus retroflexus (Table 1). Results are in accordance ences in the numbers of surviving weed plants when with Wan et al. (1992) who reported 100% efficacy the plots treated with different herbicide applica- against Echinochloa crus-galli and 60–100% control tion rates were compared, except for higher efficacy of Portulaca oleracea and Amaranthus tricolor at the of 2.0 RD. Pethoxamid showed excellent efficacy application rate 1119 g/ha. The excellent efficacy of (mostly good to very good control) on E. crus-galli dimethachlor from the lowest application rate was that is in line with Anonymous (2002), Hunt et al. also observed on Lamium purpureum (very good (2015), Jursík et al. (2015; efficacy 90–97%), and

313 Original Paper Plant Protection Science, 56, 2020 (4): 305–316

https://doi.org/10.17221/93/2020-PPS

PPDB (2020). Pethoxamid also very well controlled tion and a significantly lower yield of crops in un- Lamium purpureum, L. amplexicaule, and Veroni- treated plots as a result of competition from weeds ca persica. In contrast to the insufficient efficacy have been reported by many authors (e.g. Ritter et al. of pethoxamid on Portulaca oleracea reported by 2008; Pannacci & Onofri 2016). The application of Anonymous (2002), we found very good to full con- dimethachlor, however, caused poorer development trol of this species by this herbicide (Table 1). Using of inconsumable rocket parts in comparison with higher application rates (≥ RD), pethoxamid suf- the application of pethoxamid. ficiently or acceptably controlled Anagallis arven- In 2016, we found high rocket germination and sis and Chenopodium spp. Anonymous (2002) and survival approaching 90–100% in untreated plots PPDB (2020) reported the efficacy of pethoxamid on and a low (pethoxamid) or high decrease (dimetha- Chenopodium album, however, Jursík et al. (2015) chlor) in rocket densities in the herbicide-treated found insufficient control. Our experiments did not plots. Because the weed densities were ten times confirm the efficacy of pethoxamid Persicariaon lower in comparison with the 2015 experiment maculosa reported by PPDB (2020), although also when an effect of "dilution" of applied doses of her- according to Anonymous (2002) the control of peth- bicides among extremely high numbers of weed oxamid should be insufficient on this species. We individuals could occur, it is likely that in 2016 also found the efficacy of pethoxamid on A. retro- a higher proportion of the applied dose of herbicide flexus (Table 1) comparable with previous research left in the soil and could, therefore, have been ab- (Anonymous 2002; Hunt et al. 2015; Jursík et al. sorbed by the growing rocket plants, leading either 2015: 91–99%; Soltani et al. 2019: 98%). Control of to neutral (pethoxamid) or even disadvantageous Thlaspi arvense was low to insufficient. According (dimethachlor) effects on the germination and de- to available data, pethoxamid should be efficient also velopment of the rocket. To summarize, any herbi- on Convolvulus arvensis Linnaeus, Digitaria Haller, cide treatment tended to increase the yield but also Panicum Linnaeus, Setaria P. Beauv., and Solanum the variation in the consumable rocket parts in the Linnaeus (Hunt et al. 2015; PPDB 2020). 2015 experiment, suggesting a weak positive effect Our results showed different effects of the two her- of the herbicide treatment on the rocket yield, which bicides on the rocket yield parameters, depending on is generally expected in strongly weeded trials when the year of the experiment. Such contrasting results a threshold weed density level for crop stands is emphasize the importance of repeated testing under exceeded (Ritter et al. 2008; Gerhards et al. 2011). contrasting environmental conditions. Dimetha- However, if the weed competition is relatively low, chlor showed lower selectivity to a rocket than peth- as observed in the year 2016, the total effect of the oxamid and it, therefore, reduces the production of herbicide treatment on the yield is usually close to the rocket. In 2015, the experiment was carried out zero or negative, as in our case, as a result of its phy- in a heavily weeded field, which proved to be an ad- totoxic effect (Gerhards et al. 2011). vantage, as both herbicides could be tested on the Considering the usability of the herbicides in the efficacy on a broad spectrum of weed species and on rocket, pethoxamid seems to be more suitable, de- their density. Despite the significant effect of the her- spite its slightly lower efficacy on weeds. It follows bicides against the weeds, only 50–75% of the initial- that the survival of the rocket plants and the yield of ly sowed rocket seeds reached maturity, irrespective consumable parts were higher than, or comparable of the treatment. One possible explanation of such to, the control plots after the application of pethoxa- a contrasting result could be that the high densities of mid in any application rate that was applied, com- weeds caused a strong competitive effect early in the pared to the mostly negative effects of dimethachlor season (Gallandt & Weiner 2015; Pannacci & Onofri on the rocket yield. Moreover, even the lowest herbi- 2016) that strongly affected the growth of the rocket cide application rate used in the experiments (0.8 RD) seedlings. However, a marginally significant trend is sufficient to satisfy both the rocket yield and weed of the weight of consumable rocket parts increasing control. Though the RD of both herbicides was after the application of any tested doses of herbi- recommended for oilseed rape and not for a rocket, cides was recorded in the 2015 experiment, confirm- our data is in agreement with other studies on different ing the positive effect of herbicide treatment on the crops that demonstrate the overestimation of recom- final consumable yield of the rocket as a result of the mended doses for many herbicides (Zhang et al. 2000; removal of competition from weeds. Growth reduc- Gaba et al. 2016). Herbicides in reduced doses are

314 Plant Protection Science, 56, 2020 (4): 305–316 Original Paper https://doi.org/10.17221/93/2020-PPS often sufficient to control weed density at or below Acknowledgement: The authors thank the staff the threshold levels and thus maintain satisfactory of the Central Institute for Supervising and Testing crop yields (Steckel et al. 1990; Gaba et al. 2016). in Agriculture in Olomouc for their valuable help In food, herbicide residues are found minimally, with setting up the experiments. Authors thank two compared to insecticides or fungicides (Anony- reviewers for valuable comments on the original mous 2017; EFSA 2018). However, evaluation of manuscript. which herbicide and in which application rate is suitable for use in a certain crop in terms of harm- REFERENCES lessness to health at the stage of harvest maturity must be based on analysis of the residues of Anonymous (2002): Monograph Pethoxamid. Vol. 3. Annex herbicides in the resulting product (Keikotlhaile B-3: Data on application and further information. Available & Spanoghe 2011). The RDs used in our study were at https://www.bvl.bund.de/SharedDocs/Downloads/04_ primarily based on the application doses recom- Pflanzenschutzmittel/02_eu_berichte/Pethoxamid-DAR. mended mainly for rapeseed (Anonymous 2014), pdf?__blob=publicationFile&v=3 (accessed Jul 20, 2020). which is harvested for seed and therefore has Anonymous (2005): Regulation (EC) No. 396/2005 of the a much longer time from the application of the her- European Parliament and of the Council of 23 February bicide to harvesting to degrade the residues. In the 2005 on maximum residue levels of pesticides in or on food rocket, the dimethachlor residue was below the and feed of plant and animal origin and amending Council limit of quantification (LOQ; 0.003 mg/kg) when Directive 91/414/EEC. the RD was applied. This finding is consistent with Anonymous (2014): List of authorized plant protection the results of Šuk et al. (2018), which reported fast products and other means of plant protection 2014. Bul- degradation of dimethachlor and recommended letin of the Central Institute for Supervising and Testing in it for non-residue production and for products in- Agriculture, 1: 1, 2014. Available at http://eagri.cz/public/ tended as infant food. The toxicological evalua- app/eagriapp/POR/Files/VESTNIK_2014_LEDEN.pdf (action and environmental behaviour studies allow the cessed Mar 3, 2019). classification of pethoxamid into the safe herbicide Anonymous (2017): Pesticide Residue Monitoring Program group (Kato et al. 2001). Rocket harvested from the - Fiscal Year 2017 Pesticide Report. U.S. Food and Drug plots treated with pethoxamid using the RD had Administration. Available at https://www.fda.gov/me- weakly positive findings but all below 0.010 mg/kg dia/130291/download (accessed Mar 2, 2020). (a 'default limit'), which is the baseline MRL for food, Anonymous (2020): BS EN 15662: 2008. Foods of plant unless otherwise specified (Anonymous 2005). How- origin - Multimethod for the determination of pesticide ever, since the pethoxamid residue content in the residues using GC- and LC-based analysis following ace- leaves harvested from the 0.8 RD-treated plots did not tonitrile extraction/partitioning and clean-up by dispersive differ significantly from the untreated plots in either SPE - Modular QuEChERS-method. Available at http:// of the experimental years and the pethoxamid con- www.chromnet.net/Taiwan/QuEChERS_Dispersive_SPE/ tent was always below the limit of quantification, we QuEChERS_%E6%AD%90%E7%9B%9F%E6%96%B9%E have decided to recommend this dose, i.e., 960 g/ha 6%B3%95_EN156622008_E.pdf (accessed Jul 20, 2020). of pethoxamid, for weed control in rocket stands. Böger P., Matthes B., Schmalfuß J. (2000): Towards the pri- mary target of chloroacetamides – new findings pave the CONCLUSION way. Pest Management Science, 56: 497–508. Damalas C.A., Eleftherohorinos I.G. (2011): Pesticide expo- Pethoxamid showed no or minimal phytotoxicity to sure, safety issues, and risk assessment indicators. Inter- the rocket. Although this herbicide is slightly less ef- national Journal of Environmental Research and Public fective against weeds than dimethachlor, pethoxam- Health, 8: 1402–1419. id eliminates many weeds reliably. The dose 960 g/ha Dimson E.V. (2001): Crop Profile for Arugula in Arizona. The is safe in terms of the presence of residues in the Crop Profile/IPM database. Available athttps://cals.arizona. rocket leaves at harvest maturity. The time from edu/crop/public/docs/AZarugula.pdf (accessed Dec 10, 2019). the pre-emergent application of the herbicide to the Doležalová I., Duchoslav M., Dušek K. (2013): Biology and yield harvesting of the rocket leaves, which is about seven of rocket (Eruca sativa Mill.) under field conditions of the weeks, is sufficient for the degradation of pethoxa- Czech Republic (Central Europe). Notulae Botanicae Horti mid in the harvested product. Agrobotanici Cluj-Napoca, 41: 530–537.

315 Original Paper Plant Protection Science, 56, 2020 (4): 305–316

https://doi.org/10.17221/93/2020-PPS

Doležalová I., Koprna R., Petrželová I., Duchoslav M. (2014): The urgent need for a new concept in agriculture. Frontiers Možné způsoby chemické ochrany rokety seté – předběžná in Public Health, 4: 148. doi: 10.3389/fpubh.2016.00148 studie. Aktuální poznatky v pěstování, šlechtění, ochraně OEPP/EPPO (2007): Weeds in brassica oil crops, PP1/049(3). rostlin a zpracování produktů. Úroda, 12: 517–520. OEPP/EPPO Bulletin, 37: 36–39. EFSA (European Food Safety Authority) (2018): The 2016 Padulosi S. (1995): Rocket Genetic Resources Network. Re- European Union report on pesticide residues in food. EFSA port of the First Meeting, Lisbon, Nov 13–15, 1994: 1–70. Journal, 16: 5348. doi: 10.2903/j.efsa.2018.5348 Pannacci E., Onofri A. (2016): Alternatives to terbuthylazine Gaba S., Gabriel E., Chadœuf J., Bonneu F., Bretagnolle V. for chemical weed control in maize. Communication in (2016): Herbicides do not ensure for higher wheat yield, but Biometry Crop Science, 11: 51–63. eliminate rare plant species. Scientific Reports, 6: 30112. PPDB (2020): The Pesticide Properties Database (PPDB) de- doi: 10.1038/srep30112 veloped by the Agriculture & Environment Research Unit Gallandt E.R., Weiner J. (2015): Crop-weed competition. (AERU), University of Hertfordshire. Available at http://sitem. Encyclopedia of Life Sciences, 2nd ed. Chichester, John herts.ac.uk/aeru/ppdb/en/index.htm (accessed Jul 20, 2020). Wiley & Sons, Ltd. Ritter C., Dicke D., Weis M., Oebel H., Piepho H.P., Büchse Gerhards R., Gutjahr C., Weis M., Keller M., Sökefeld M., A., Gerhards R. (2008): An on-farm approach to quantify Möhring J., Piepho H.P. (2011): Using precision farming yield variation and to derive decision rules for site-specific technology to quantify yield effects attributed to weed com- weed management. Precision Agriculture, 9: 133–146. petition and herbicide application. Weed Research, 52: 6–15. Soltani N., Brown L.R., Sikkema P.H. (2019): Weed Control Hensel D.R. (2005): Nondetects and Data Analysis: Statistics for in Corn and Soybean with Group 15 (VLCFA Inhibitor) Censored Environmental Data. Hoboken, John Wiley & Sons. Herbicides Applied Preemergence. International Journal Holmstrom K.E. (2008): Crop profile for Arugula in New Jer- of Agronomy, 2019: 1–7. sey. National IPM Database. Available at https://ipmdata. Steckel L.E., DeFelice M.S., Sims B.D. (1990): Integrating ipmcenters.org/documents/cropprofiles/NJarugula.pdf reduced rates of postemergence herbicides and cultivation (accessed Dec 15, 2019). for broadleaf weed control in soybeans (Glycine max). HRAC (2010): The World of Herbicides. According to HRAC Weed Science, 38: 541–545. classification on mode of action 2010. Available at https:// Šmilauer P., Lepš J. (2014): Multivariate Analysis of Eco- hracglobal.com/files/moaposter.pdf (accessed Nov 19, 2019). logical Data Using. Canoco 5. Cambridge, Cambridge Hunt B., Barrentine J., Hayden T. et al. (2015): Pethoxamid University Press. — a new herbicide for use in agronomic and horticultural Šuk J., Jursík M., Suchanová M., Schusterová D., Hamouzová crops. In: Proceedings of 55th Annual Meeting of Weed Sci- K. (2018): Dynamics of herbicide degradation in cauli- ence Society of America, Lexington, Feb 9–12, 2015: 281. flower. Plant Soil and Environment, 64: 551–556. Jursík M., Soukup J., Holec J., Andr J., Hamouzová K. (2015): ter Braak C.J.F., Šmilauer P. (2012): Canoco Reference Manual Efficacy and selectivity of pre-emergent sunflower -her and User’s Guide: Software for Ordination (version 5.0). bicides under different soil moisture conditions. Plant Ithaca, Microcomputer Power. Protection Science, 51: 214–222. Tomášek M. (2015): Půdy České republiky. Praha, Česká Kato S., Kitajima T., Okamoto H., Kobutani T. (2001): geologická služba. Pethoxamid – a novel selective herbicide for maize and Wan C.X., Leng D.X., Ma S.N., Dong Y.H. (1992): Technique for soybean. In: Brighton Crop Protection Conference Weeds, the control of weeds in plastic film mulched groundnut fields 1, British Crop Protection Council, Brighton, Nov 12–15, with herbicide dimethachlor. Plant Protection, 18: 50–51. 2001: 23–28. Yang X., Guschina I.A., Hurst S., Wood S., Langford M., Keikotlhaile B.M., Spanoghe P. (2011): Pesticide residues in Hawkes T., Harwood J.L. (2010): The action of herbicides fruits and vegetables. In: Stoytcheva, M. (ed.): Pesticides: on fatty acid biosynthesis and elongation in barley and Formulations, Effects, Fate. Rijeka, InTech: 243–252. cucumber. Pest Management Science, 66: 794–800. Legendre P., Legendre, L. (2012): Numerical Ecology. 3rd Zhang J., Weaver S.E., Hamill A.S. (2000): Risks and Reli- English ed. Amsterdam, Elsevier Science BV. ability of Using Herbicides at Below-Labeled Rates. Weed Lepš J., Šmilauer P. (2014): Multivariate Analysis of Ecological Technology, 14: 106–115. Data using Canoco. 2nd Ed. Cambridge, Cambridge Uni- versity Press. Nicolopoulou-Stamati P., Maipas S., Kotampasi C., Stamatis Received: June 25, 2020 P., Hens L. (2016): Chemical pesticides and human health: Accepted: August 22, 2020 Published online: September 18, 2020

316